pubcrawl

PubCrawl is an MCP server that exposes a curated set of tools to access biomedical literature, drug labelling information, and ClinicalTrials data. It aggregates PubMed search and retrieval (citations, abstracts, full text where available), US and UK drug labeling, and ClinicalTrials.gov records, making it possible to query and format results for downstream LLM interactions. Tools include search_pubmed, get_abstract, get_full_text, find_related, format_citation, trending_papers, get_uspi, get_smpc, compare_labels, search_by_indication, search_trials, and get_trial. To use it, install the pubcrawl CLI or run it from your environment as configured, then issue structured prompts that target one of the tools (for example, asking for a PubMed search, retrieving an abstract, or comparing US and UK labelling). The server handles the underlying API calls and returns structured results, enabling your LLM client to present concise summaries, citations, and regulatory information with provenance where available.

Prerequisites:

• Node.js 20+ installed on your system
• Access to a terminal or shell

Option A: Install via npm (global CLI)

1. Install the package globally so you can run the pubcrawl command from anywhere:

npm install -g @pharmatools/pubcrawl

2. Run the server or connect to it from your MCP client by pointing to the pubcrawl command:

# Example: start the server (if the CLI supports running as a server)
pubcrawl

Option B: Install from source and build

1. Clone the repository and install dependencies:

git clone https://github.com/nickjlamb/pubcrawl.git
cd pubcrawl
npm install

2. Build the project to generate distributable artifacts:

npm run build

3. Point your MCP config to the built file, for example:

{
  "mcpServers": {
    "pubcrawl": {
      "command": "node",
      "args": ["/path/to/pubcrawl/dist/index.js"]
    }
  }
}

Option C: Use via npx (on-demand, no global install)

1. If you prefer not to install globally, you can run via npx:

npx -y @pharmatools/pubcrawl

Note: If you use an API key (for NCBI or other services), ensure you configure the environment variables as described in the README and pass them via your MCP configuration when starting the server.

Tips and considerations:

• If you plan to make a high volume of PubMed or ClinicalTrials requests, consider obtaining and configuring API keys (e.g., NCBI API Key) to increase rate limits.
• The PubCrawl tools return structured data (PMIDs, DOIs, trial IDs, etc.) suitable for prompting LLMs to generate formatted citations or summaries.
• Ensure your MCP client supports the specific tool invocation names (e.g., search_pubmed, get_abstract) and that the server endpoint is reachable from your environment.
• For UK Labelling, ensure access to eMC/SmPC data if you intend to compare USPI vs SmPC outputs; some data sources may require you to respect terms of use.
• If you encounter rate limits or connectivity issues, verify network access to external resources (PubMed, FDA, eMC, ClinicalTrials.gov) and check any API key configuration in your environment variables.